Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
20th January 2025, 01:09 | #21 | Link | |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
Quote:
Next time please run your tests with a simple BlankClip(), or even better, with a "REAL-WORLD" video file. UPDATE: The problem is not in ColorBarsHD() itself or its default colorspace YV24, it's the "preferred resizing dimensions" chosen by DTL (2000x2000) — for some obscure reason, JincResize does not like them sometimes.
__________________
«Your software patents have expired.» Last edited by filler56789; 20th January 2025 at 04:58. Reason: add more info |
|
20th January 2025, 18:32 | #22 | Link |
Registered User
Join Date: Jul 2018
Posts: 1,230
|
Random crashes with memory access violation are strange and harder to debug (longer time to reproduce at least to see what happen with addresses and attempt to understand where is it come from). If it somehow around internal multithreading - try to set threads=1 to disable internal multithreading and look if the crash still happen (at some runs).
Addition: Testbuild of x86 version with about 3x lower RAM usage for 4:4:4 and RGB formats added https://github.com/Asd-g/AviSynth-Ji...ent-2603071285 . Though it runs with Jinc256Resize(3200,2000) from AVSmeter with only about 600 MiB RAM usage - attempt to open same script in VirtualDub still fail with memory allocation (formatted) error. Last edited by DTL; 20th January 2025 at 20:06. |
21st January 2025, 01:19 | #23 | Link | |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
Quote:
Okay, here we go again. 1) SCRIPT Code:
LoadPlugin("C:\AVSplugins32\JincResize-LAST02.dll") ColorBarsHD() KillAudio() Code:
Jinc256Resize(3840,2160) ##### OK Jinc256Resize(640,360) #### OK Jinc256Resize(2000,2000) #JincResize: failed to allocate memory for coefficient buffer. Jinc144Resize(1000,1000) # OK Jinc256Resize(1000,1000) # OK Jinc256Resize(3000,3000) #VirtualDub crashes Jinc256Resize(4000,2250) #OK Jinc256Resize(1600,1600) #OK Jinc256Resize(2400,2400) #OK Jinc256Resize(8000,4500) #JincResize: failed to allocate memory for coefficient buffer. JincResize(8000,4500) #Opens OK, but VirtualDub crashes during playback. Jinc36Resize(8000,4500) #Opens OK, but VirtualDub crashes during playback. Jinc64Resize(8000,4500) #VirtualDub crashed Jinc144Resize(8000,4500) #JincResize: failed to allocate memory for coefficient buffer. ¿what is so special in the video resolution 2000x2000? ¿Does Asd-g's JincResize dislike it by design?
__________________
«Your software patents have expired.» Last edited by filler56789; 21st January 2025 at 01:22. Reason: grammar |
|
21st January 2025, 06:09 | #24 | Link |
Registered User
Join Date: Jul 2018
Posts: 1,230
|
"JincResize(8000,4500) #Opens OK, but VirtualDub crashes during playback.
Jinc36Resize(8000,4500) #Opens OK, but VirtualDub crashes during playback." Thank you. This is more easy way to finally catch the crash in the debugger and start looking what is wrong. Crash after OK loading (processing) of some frames is more strange because addresses/memory allocations are not changes at GetFrame() and sort of nothing changes. Only random come from threads switching (multithreading and normal threads switch). When you got crash at playback - at the first frames or after some 100..1000+ ? The 8000x4500 runs very slow at my old CPU but still can not got crash at playback (about 20 frames tested). Last edited by DTL; 21st January 2025 at 06:20. |
21st January 2025, 14:23 | #25 | Link | |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
Quote:
Still, you haven't answered my last question. ¿WHY Jinc256Resize happily accepts 4000x2250 OR 3840x2160 BUT "fails to allocate memory" for 2000x2000? 4000x2250 > 2000x2000; 3840x2160 > 2000x2000;
__________________
«Your software patents have expired.» Last edited by filler56789; 21st January 2025 at 14:34. Reason: MORE clarity |
|
21st January 2025, 17:38 | #26 | Link |
Registered User
Join Date: Jul 2018
Posts: 1,230
|
"WHY Jinc256Resize happily accepts 4000x2250 OR 3840x2160 BUT "fails to allocate memory" for 2000x2000? "
Do not have ideas now. My latest tests is that Jinc144Resize(3000,3000) about stable crash in release build at opening in VirtualDub and no one time crash in debug build and with some frames resize in 'playback' in VirtualDub when running from VS debugger. Sad - can not easy catch the crash in the debugger. The release build even with debug info with optimizations enabled not very good match the source. So the debug of some annoying crashes may take many time. |
23rd January 2025, 04:35 | #27 | Link |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
JincResize v2.1.3 has been released!
The problem, of course, is Asd-g still is too lazy to build a 32-bit DLL https://github.com/Asd-g/AviSynth-Ji...ases/tag/2.1.3
__________________
«Your software patents have expired.» |
23rd January 2025, 05:25 | #28 | Link |
Registered User
Join Date: Oct 2024
Posts: 3
|
You can find 32-bit build here: https://github.com/Asd-g/AviSynth-Ji...ent-2608657460
If you not have found it yet. |
23rd January 2025, 05:40 | #29 | Link | |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
Quote:
2) Still, that is no excuse for a) not including the DLL in the "official" archive and b) keep saying the 32-bit edition is "unsupported" *Not recommended*, OK, I agree with this. But "unsupported", this makes no sense. Because after all, it was because the 32-bit DLL was compiled and tested that both DTL and Asd-g could make the source-code less-inefficient
__________________
«Your software patents have expired.» |
|
23rd January 2025, 14:00 | #30 | Link |
Registered User
Join Date: Jul 2018
Posts: 1,230
|
32bit version still very frequently crashes at attempt to run simple script
LoadPlugin("JincResize.dll") ColorBarsHD() Jinc256Resize(3000,3000) And I think 'unsupported' means Asd-g does not like to search why these (non-repeatable in a sequence) crashes happen in release 32bit build. So it may left as 'hanging issue' for unknown time (or simply ignored after 32bit machines will die in the past completely). Last edited by DTL; 23rd January 2025 at 14:02. |
24th January 2025, 13:08 | #31 | Link |
Registered User
Join Date: Jul 2018
Posts: 1,230
|
One more test version with attempt to fix crashes - https://github.com/Asd-g/AviSynth-Ji...ent-2612169256
|
Yesterday, 04:51 | #32 | Link | |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
Quote:
Well, it SEEMS the problem with 2000x2000 is gone. BUT the trouble with 3000x3000 remains. And it remains as "illogical" as before. I mean, Jinc256Resize(4096,2304) opens fine and plays fine, whereas Jinc256Resize(3000,3000) still makes VirtualDub crash, —even though 3000x3000 < 4096x2304. 🤔
__________________
«Your software patents have expired.» |
|
Yesterday, 14:44 | #34 | Link |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,441
|
Good idea.
Below is the content of crashinfo.txt. The problem was, the "modern" versions of Windows are exceedingly meddlesome and love to get in the way (I mean, I had to find a way to fool the annoying "VirtualDub has stopped working" pop-up generated by the operating system itself). Code:
VirtualDub crash report -- build 35491 (release) -------------------------------------- Disassembly: 733f6e40: 0f8e0e010000 jle 733f6f54 733f6e46: 8b4610 mov eax, [esi+10h] 733f6e49: c5d0 lds edx, eax 733f6e4b: 57 push edi 733f6e4c: ed in eax, dx 733f6e4d: c5d9 lds ebx, ecx 733f6e4f: ef out dx, eax 733f6e50: e48b in al, 8bh 733f6e52: 088b4618894c or [ebx+4c891846], cl 733f6e58: 240c and al, 0ch 733f6e5a: 8b38 mov edi, [eax] 733f6e5c: 8b4614 mov eax, [esi+14h] 733f6e5f: 897c2420 mov [esp+20h], edi 733f6e63: 8b30 mov esi, [eax] 733f6e65: 8b44241c mov eax, [esp+1ch] 733f6e69: 0fafc2 imul eax, edx 733f6e6c: 8974242c mov [esp+2ch], esi 733f6e70: 8d0440 lea eax, [eax+eax*2] 733f6e73: c1e002 shl eax, 02h 733f6e76: 89442418 mov [esp+18h], eax 733f6e7a: 660f db 0fh 733f6e7c: 1f pop ds 733f6e7d: 44 inc esp 733f6e7e: 0000 add [eax], al 733f6e80: 8b4904 mov ecx, [ecx+04h] 733f6e83: c5fc lds edi, esp 733f6e85: 10dd adc ch, bl 733f6e87: 8b540104 mov edx, [ecx+eax+04h] 733f6e8b: 0fafd7 imul edx, edi 733f6e8e: 031401 add edx, [ecx+eax] 733f6e91: 8b4c0108 mov ecx, [ecx+eax+08h] 733f6e95: 03d6 add edx, esi 733f6e97: 8b44240c mov eax, [esp+0ch] 733f6e9b: 8b00 mov eax, [eax] 733f6e9d: 8d3488 lea esi, [eax+ecx*4] 733f6ea0: 8b4c240c mov ecx, [esp+0ch] 733f6ea4: 8b790c mov edi, [ecx+0ch] 733f6ea7: 85ff test edi, edi 733f6ea9: 7e59 jle 733f6f04 733f6eab: 8b4110 mov eax, [ecx+10h] 733f6eae: c1e002 shl eax, 02h 733f6eb1: 89442424 mov [esp+24h], eax 733f6eb5: 897c2410 mov [esp+10h], edi 733f6eb9: 0f db 0fh 733f6eba: 1f pop ds 733f6ebb: 800000 add byte ptr [eax], 00h 733f6ebe: 0000 add [eax], al 733f6ec0: 33c0 xor eax, eax 733f6ec2: 8bce mov ecx, esi 733f6ec4: 0f db 0fh 733f6ec5: 1f pop ds 733f6ec6: 40 inc eax 733f6ec7: 000f add [edi], cl 733f6ec9: 1f pop ds 733f6eca: 8400 test [eax], al 733f6ecc: 0000 add [eax], al 733f6ece: 0000 add [eax], al 733f6ed0: c4e2 les esp, edx 733f6ed2: 7d31 jge 733f6f05 733f6ed4: 0410 add al, 10h 733f6ed6: c5fc lds edi, esp 733f6ed8: 10d3 adc bl, dl 733f6eda: c5fc lds edi, esp 733f6edc: 5b pop ebx 733f6edd: d8c4 fadd st, st(4) 733f6edf: e26d loop 733f6f4e 733f6ee1: 98 cwde 733f6ee2: 1983c0088d49 sbb [ebx+498d08c0], eax 733f6ee8: 20c5 and ch, al 733f6eea: fc cld 733f6eeb: 10c3 adc bl, al 733f6eed: 3bc7 cmp eax, edi 733f6eef: 7cdf jl 733f6ed0 733f6ef1: 03742424 add esi, [esp+24h] 733f6ef5: 03542420 add edx, [esp+20h] 733f6ef9: 836c241001 sub dword ptr [esp+10h], 01h 733f6efe: 75c0 jnz 733f6ec0 733f6f00: 8b4c240c mov ecx, [esp+0ch] 733f6f04: 8b542414 mov edx, [esp+14h] 733f6f08: 8b742428 mov esi, [esp+28h] 733f6f0c: 8b7c2420 mov edi, [esp+20h] 733f6f10: c4e3 les esp, ebx 733f6f12: 7d19 jge 733f6f2d 733f6f14: d801 fadd dword ptr [ecx] 733f6f16: c5f8 lds edi, eax 733f6f18: 58 pop eax 733f6f19: c3 ret 733f6f1a: c5fb lds edi, ebx 733f6f1c: 7cc0 jl 733f6ede 733f6f1e: c5fb lds edi, ebx 733f6f20: 7cc0 jl 733f6ee2 733f6f22: c5f9 lds edi, ecx 733f6f24: 5b pop ebx 733f6f25: c0c4e2 rol ah, 0e2h 733f6f28: 792b jns 733f6f55 733f6f2a: c4c5 les eax, ebp 733f6f2c: f9 stc 733f6f2d: 67c4c5 les eax, ebp 733f6f30: f9 stc 733f6f31: 7ec0 jle 733f6ef3 733f6f33: 880432 mov [edx+esi], al 733f6f36: 42 inc edx 733f6f37: 8b442418 mov eax, [esp+18h] 733f6f3b: 8b74242c mov esi, [esp+2ch] 733f6f3f: 83 db 83h Built on Althena on Sun Oct 27 15:59:07 2013 using compiler version 1400 Windows 6.3 (Windows 7 x64 build 9600) [] Memory status: virtual free 1667M/2048M, commit limit 10125M, physical total 8077M EAX = 00000000 EBX = 008dce00 ECX = 0d46c000 EDX = 04735ed5 EBP = 0f18fd8c ESI = 0d46c000 EDI = 00000011 ESP = 0f18fd58 EIP = 733f6ede EFLAGS = 00010246 FPUCW = 027f FPUTW = ffff Crash reason: Access Violation Crash context: An out-of-bounds memory access (access violation) occurred in module 'JincResize'... ...reading address 0D46C000. Pointer dumps: EBX 008dce00: 00000060 00000061 00000062 00000063 00000064 00000065 00000066 00000067 EDX 04735ed1: 68686868 68686868 68686868 68686868 68686868 68686868 68686868 68686868 ESP 0f18fd58: 008dcdd4 0018f4c0 0f18fe2c 00888cb8 0000000b 00000045 002eb45c 00000bb8 0f18fd78: 00000540 00000080 11ccf700 04730040 00000001 0f18fda8 733f92e9 00000055 0f18fd98: 0086faf0 733f9420 0b110f58 0018f468 0f18fde8 733f9429 77253677 0f18fe2c 0f18fdb8: 0018f468 0086faf0 0080b830 0086fb50 77253540 00000000 0b110f58 00000000 EBP 0f18fd88: 00000001 0f18fda8 733f92e9 00000055 0086faf0 733f9420 0b110f58 0018f468 0f18fda8: 0f18fde8 733f9429 77253677 0f18fe2c 0018f468 0086faf0 0080b830 0086fb50 0f18fdc8: 77253540 00000000 0b110f58 00000000 0086faf0 0b110f28 00000018 26b7ab76 0f18fde8: 0f18ff80 77250e61 0f18fe2c 0086fb50 26b7a93a 77244870 77244870 0080b830 Thread call stack: 733f6ede: JincResize!_avisynth_c_plugin_init@4 [733f0000+3860+367e] 733f92e9: JincResize!_avisynth_c_plugin_init@4 [733f0000+3860+5a89] 733f9429: JincResize!_avi
__________________
«Your software patents have expired.» |
Thread Tools | Search this Thread |
Display Modes | |
|
|